Identifying the optimum production strategy for a field development plan is a critical decision to be made with reservoir management. Drilling of wells is costly, often amounting to hundreds of millions of dollars for a reservoir, depending on the number and types of wells involved. Once wells are drilled, there is no flexibility in changing surface locations later. The only possible alteration is to call workover rigs to sidetrack or recomplete the low profile wells, but the surface location will remain unchanged. Typically, the selection process for locating wells utilizes available technology such as saturation applications, pressure distribution maps, isopach maps, rock quality maps, and so forth. The selection process also employs, to some degree, experience and intuition. The current worldwide decline in new hydrocarbon reservoir discoveries dictates that locating and placing wells be more justified by reducing the intuition involved and including robust optimization process to improve the quality of wells placement for optimal recovery.
Assessing wells placement is usually a twostep process. The first part of the process is to identify sweet spots in a reservoir. The second step includes the optimization formulation or method to be used. Several methods are known to identify sweet spots in a reservoir. One known technique is Genetic Unit (GU) which designates distinctive parts of the reservoir rock based on static properties such as geometrical, petrophysical, and spatial properties. The objective using GU is to optimize horizontal or high angle well trajectories, which are defined by azimuth, location, inclination, and length, in order to maximize productivity and drainage by intersecting high permeability GU's. The ultimate goal is to develop trajectories that interconnect with a larger number of productive GU's and rank them based on their values and the associated risk with each trajectory. This method utilizes static data solely, and the ranking criterion used are objective.
Computer programs have also been written to calculate three dimensional geo-object properties such as porosity, permeability, and lithofacies. One means that can be used as a net indicator is to represent each cell in a model which can be ranked according to its connectivity and then used to place wells. This has been extended by incorporating geologic uncertainty with an optimization set-up where the objective is to maximize the connected geo-objects that are meant to intersect with a minimum number of planned wells. For the optimization loop, a simulated annealing algorithm is used to find the best location for vertical, deviated, and horizontal wells.
Another approach computes two-dimensional indicators, referred to as reservoir quality, for each cell in a reservoir model. The quality term represents connected net pay modified by tortuosity. Binary integer programming can be used to find the best well sites with the goal of maximizing reservoir quality and incorporating minimum spacing between wells. This work has been extended into three-dimensional models in which the combination of quality indicators and geo-objects is used to identify optimum completions.
Another method has been suggested in which a productivity proxy function is created attempting to account for possible attributes by compacting grid blocks in a reservoir model. The productivity potential map accounts for porosity, permeability, and oil saturation. This method can also join preferred phase pressure and distance to the nearest boundary to the productivity proxy function to guide wells placement strategy in bottom water drive reservoirs.
A further method has identified multiple well targets to be maximized by utilizing two categories of filters. The first is the contact between the wells and the ranked pore volume measures in certain drainage radius. The other category includes permeability, distance from fluid contacts, and saturation. The resulting targets accommodate spacing constraints before establishing well paths or plans that follow certain designs for different types of wells. The goal of this method is to find the optimum location and types for production or injection wells by maximizing total hydrocarbon production or net present value. The method links optimization scheme with a reservoir stimulator and allows limited movements for each well type in order to achieve the highest objective function.
These methods discussed use mainly static parameters. Static parameter utilization, however, is not sufficient to find the optimum location of wells. In fact, once the field starts producing, parameters will start changing due to fluid flow, change in fluid saturation, and so forth. Therefore, it is accepted by the industry that dynamic maps generated from the dynamic parameters provide accurate images showing the quality of a reservoir rock relative to the active forces within.
The other category of sweet spots delineation is based on dynamic measures. Quality map term is a form of analysis where a single vertical well is placed in different area grid locations in a model to create two-dimensional cumulative oil maps. Ideally, the model is run for every well on several grid locations for a sufficient amount of time to understand trends in collective oil volume produced. This approach can be used to locate wells with an optimization algorithm under geological uncertainty. Later, the quality map approach can be used to generate quality maps for oil, gas, and water production to improve the initial population in wells placement optimization processes using genetic algorithm. The best producer locations are identified using oil and gas quality maps, and the beta injector locations are determined using water quality maps, which yield the maximization of either cumulative hydrocarbon or net present value.
Others have used the productivity index as an objective function to be maximized by optimum well locations. Numerical productivity index and field productivity index have been suggested as two different attributes to be maximized. The values of the numerical productivity index can be obtained from the production of horizontal wells in a reservoir model. A field productivity index can be obtained by the calculation of flow rate over total drawdown for one large time step. These attributes can be used with a variance matrix adaptation algorithm to find the optimum wells placement.
The quality maps approach has enticed many researchers and has been used more often when compared with other methods. The fact that it requires long simulation runs to obtain one map, especially in large models, could cause researchers to shy away from implementing quality maps approach in real reservoirs.